Think About It. Telephone Poll? Professional psychics gain their reputations by being correct once in a while. Even the quatrains of Nostradamus contain cloudy contradictions perfectly fit for future interpreters seeking fame of their own. Son Jim, a good young magician, a skilled prestidigitator, also found a book on “cold reading” which had great tips for describing the past and future of subjects much his senior. Northern Indiana mayors were naturally amazed at his apparent prescience. Why comment on pseudo psychics when writing an essay on polling”? You have probably noticed that the science of intention sampling is faulty at best and often fraudulent (intentionally swaying opinion, as in so-called “push polls”). The new Showtime seer series featuring 21-year old Tyler Henry, with practice and style could be based on the first paragraph of Cold reading, Basic procedure from Wikipedia.
A few months ago I made the note where I keep “thought starters”, intended as a reminder to consider the accuracy of surveys reaching respondents on their traditional home phones. Is it possible to estimate the future conduct of a general population by extrapolating from a few of these communications? The present proliferating industry is an outgrowth of the old Hooper ratings replaced by Nielsen, now Arbitron in measuring media. Now we are bombarded by seemingly innumerable attempts at reporting public opinion with a view to predicting future voter conduct. Things appear to go smoothly until a dramatic error shocks followers. In 1948, The Chicago Daily Tribune printed “DEWEY DEFEATS TRUMAN” 
as its post-election headline. Turns out that polling winner Thomas A. Dewey had lost to incumbent Harry S. Truman. In Iowa last week Ted Cruz won the Republican Presidential caucus while in-person interviews showed Donald Trump a 7 point leader. Why? Theories abound.
“Margin of error” is a good excuse when you’ve sampled 10,000 people; even better with only 300. Here’s one polling industry explanation. “The margin of error is usually defined as the radius of a confidence interval for a particular statistic from a survey.” Feel better now? Averaging polls is sometimes used to reduce error in individual sampling, also saving the cost of actually doing the polling. Then, of course, you can always blame shame. They even have names for the problems these poll glitchers cause. Both involve voter’s fear of being shamed by pollsters. The Bradley effect resulted from the 1982 race for governor of California, in which former Los Angeles mayor Tom Bradley (an African American) was clearly predicted to win even through exit polling. It appears respondents did not want to appear racist. Bradley lost. In Canada it’s the Shy Tory who reportedly doesn’t want to admit the socially unacceptable Conservative view of politics. Various shades of this ploy help produce faulty results.
If face-to-face interviews are often in dramatic error, do you really believe you can predict the behavior of a disparate group of individuals from a Telephone Poll? Think About It.